Hi,
just as a sidenote: apel does not support condor/lcgcondor (yet?). There
are a still a few sites using the condor jobmanager (and as there is an
lcgcondor version, I guess it counts as a 'supported' JM type, whatever
that means), and at those sites apel will not produce anything, and
consequently will fail the test. Now why sometimes it only gives a warning
instead of a failure is beyond me.
For the future: does anyone know about plans for extending apel support
for other jobmanagers? Accounting seems to be importanat (and will be
essential at some point), therefore it would be nice to know wether we
will be forced to change to pbs sometime in the future.
Cheers,
Szabolcs
On Fri, 14 Oct 2005, Yves Coppens wrote:
> Hi,
>
> It seems that quite a few sites fail the apel sft with:
>
> No Apel result set - FAILED!
>
> When I run apel manually, I get the following error message:
>
> Fri Oct 14 09:12:15 UTC 2005: apel-publisher - WARNING - Received a 'null'
> result set while querying the 'LcgRecords' table using rgma, this probably
> means the GOC is currently off-line, will therefore cancel attempt to
> re-publish
>
> My latest MeasurementDate and MeasurementTime in the LcgRecords on my mon
> are:
>
> | 2005-10-14 | 09:12:15 |
>
> So, locally things seem to work. Is it possible that the 'GOC is currently
> off-line' or that they are some problems with lcgic01?
>
> Yves
>
> On Fri, 14 Oct 2005, Kostas Georgakopoulos wrote:
>
>> Hi,
>>
>> On some occasions apel tests fail (site is GR-02-UoM) on the sft's
>> with cause:
>>
>> Checking if LcgRecords tuples are present for the site (max 3 day old):
>>
>> ++ rgma -c 'set query latest' -c 'set timeout 120' -c 'select count(*) from LcgRecords where ExecutingCE='\''alexander.it.uom.gr'\'' and MeasurementDate>='\''2005-10-11'\'''
>> ++ grep '^|'
>> ++ cut -f 2 -d ' '
>> ++ tail -n 1
>> + count=
>> + set +x
>>
>> Count:
>> No Apel result set - FAILED!
>>
>> or
>>
>> Checking if LcgRecords latest archiver is accessible:
>>
>> + rgma -c 'select count(*) from LcgRecords'
>> + grep 'Rows in set'
>> + set +x
>> LcgRecords latest archiver seems to be down. Exiting with warning.
>>
>>
>>
>> The thing is that this doesn't happen on every test. E.g it might fail
>> on one test and complete successfully on the next.
>> We haven't change anything on the site for some time now.
>> Is this a normal behavior or we have to check for something?
>>
>> Best regards
>> Kostas Georgakopoulos - University of Macedonia
>>
>
|